1. Technical Field
The present invention relates generally to an object-tracking apparatus and method in the environment of multiple non-overlapping cameras. More particularly, the present invention relates to an object-tracking apparatus and method, which are capable of tracking an identical object using color modeling without requiring a training phase in the environment of multiple non-overlapping cameras.
2. Description of the Related Art
Objects present in an image collected by a camera can be represented by three (Red, Green, Blue: RGB) color channels. Further, research into intentions to track an identical object in the environment of multiple non-overlapping cameras using a color histogram consisting of RGB color channels has been conducted.
In more detail, if it is assumed that images respectively captured by cameras Ci and Cj for an identical object are Vi and Vj, a model function, a mean Bright Transfer Function (BTF), a cumulative BTF, etc. have been presented as methods of establishing the following Equation (1):Color histogram of Vi=Color histogram of Vj(that is, Vi=f(Vj))  (1)
However, the conventional technology that has been presented requires a training phase before the application thereof. That is, in order to define a function f, a preliminary arrangement task (for example: calibration) using training data is required. Further, assuming that V1=fa(V2) and V2=fb(V3) are satisfied, and V1=fa(fb(V3)) is established, in the case where n cameras for collecting images are present, n−1 functions f are required. Further, when surrounding optical characteristics (light, illumination, etc.) change after cameras have been installed, a retraining phase for recalculating and re-generating functions f is required.
Further, when significant changes in the surrounding optical characteristics (light, illumination, etc.) of installed cameras frequently occur, differences between optical characteristics in camera views are significant, so that differences between color histograms for an identical object according to the location increase, thus causing the problem of making it impossible to use the conventional technology.
That is, even color histograms for an identical object may be displayed differently depending on changes in an external environment, in particular, light, illumination, etc. Examples of this case may include cases where cameras are installed in places having different indoor illumination brightness, where cameras are installed in an outdoor area in which changes in light are significant, where an object located in a bright place moves to a shadow area, and other cases.
Therefore, at the present time, there is required the development of object-tracking technology in the environment of multiple non-overlapping cameras, which can solve the above problems and does not require a training phase.